Stroke is a major cause of death and disability in end stage renal disease (ESRD) population on dialysis. Chronic atrial fibrillation (AF), common in the dialysis population, is a potentially-modifiable stroke risk factor. Warfarin, a well-studied intervention used to reduce stroke risk in the general population with AF, is considered appropriate therapy for dialysis patients with AF by the National Kidney Foundation. However, use rates of warfarin appear to be low even in suitable dialysis patients, which probably originates from a deep sense of skepticism among nephrologists that the results of randomized controlled trials of warfarin in the non-dialysis population are applicable to generally frail dialysis patients. Many critical issues regarding both stroke and warfarin use in dialysis patients with AF require urgent examination. An understanding of how warfarin is used in the setting of stroke risk is imperative, since bedside practitioners, guideline-forming thought leaders, and healthcare policymakers are operating in an information vacuum. Patterns of warfarin use for stroke prevention in AF have never been reported in a large sample of U.S. dialysis patients, primarily because the tools required to link longitudinal medication prescriptions to medical claims-based diagnoses at the level of the individual patient have never before been widely available. This proposal seeks to determine how demographic factors, comorbidities, and healthcare system factors (such as degree of urbanicity and the profit status of dialysis units) are associated with warfarin prescription in the dialysis population. Additionally, whether foundational healthcare system factors are associated with variation in stroke rates in this population will be examined. These questions will be investigated through the use of observational data, which offers the opportunity to efficiently undertake a large study of determinants of warfarin use in dialysis patients with AF. To accomplish these aims, a novel linkage between the United States Renal Data System (USRDS) Medicare data and Medicaid medication claims data will be used. Ultimately, the K23 project proposed here is designed to develop the applicant into a successful independent health services researcher who can investigate issues of stroke and cardiovascular disease prevention in ESRD, ultimately contributing to the reduction of disease burden in this growing population.